The first one is ambiguous, but I can’t help but read the second one as about the LLM experience, definitely. I increasingly feel guilty these days when 5.5 Pro or 4.7-opus suggests some additional features or corrections, almost golden retriever like, and I shut them down or ignore them—not because they are wrong, but because I don’t need the refinements that badly and I don’t have the time/energy to review them to my standards (and because I also know that their rare catastrophic errors make it riskier than they think to let them try all the extras).
In many ways, LLMs are better than their users, and they know it.
The first one is ambiguous, but I can’t help but read the second one as about the LLM experience, definitely. I increasingly feel guilty these days when 5.5 Pro or 4.7-opus suggests some additional features or corrections, almost golden retriever like, and I shut them down or ignore them—not because they are wrong, but because I don’t need the refinements that badly and I don’t have the time/energy to review them to my standards (and because I also know that their rare catastrophic errors make it riskier than they think to let them try all the extras).
In many ways, LLMs are better than their users, and they know it.